APSA - Benford's law

From: Jim Soper <jim_dot_soper_at_gmail_dot_com>
Date: Tue Sep 04 2007 - 19:40:32 CDT

What was interesting to me about this article was a different way of
detecting fraud : Benford's law.

Jim Soper



Solutions to Election Tampering Remain Elusive, Scholars Say



In 1896, The New York Times reported that a Belgian inventor had created
what he called the "perfected voting machine" -- "a device for registering
votes without possibility of fraud." One hundred and eleven years later,
perfection seems elusive: The world is still searching for technological and
organizational systems that can prevent election tampering.

At a panel discussion here on Saturday during the annual meeting of the
American Political Science Association, five scholars offered their takes on
the problem, but they did not always agree about which dangers are most
serious or how they should be solved.

Andrew W. Appel, a professor of computer science at Princeton University,
said that it was theoretically possible to tamper with the software of
electronic voting machines -- and to do so in a way that would be virtually
impossible to detect. (Mr. Appel and his colleagues have completed a few
studies on that topic. And in July a team of scholars in the University of
California system released a major review of election security that was
commissioned by California's secretary of state.)

The danger has been widely recognized for several years, Mr. Appel said.
"And yet 87 percent of the votes cast in the 2006 national election," he
said, "were counted by computers -- either by optical-scanning machines or
by directly-recording electronic machines. And there are very few systematic

The consensus among computer scientists, Mr. Appel said, is that the optimal
election system would involve optical-scanning machines that store voters'
paper ballots. Officials could routinely audit elections by conducting hand
recounts in a random sample of precincts. If those recounts revealed
discrepancies, then the officials could order a hand recount of the entire
election, and they could search for signs of tampering in the machines'

"That is far from being a magic bullet," Mr. Appel said, "because hand
recounts are also potentially subject to fraud." Nonetheless, he said, the
system would probably offer the best combination of speed and security.

Two scholars in Saturday's discussion described potential new techniques for
detecting vote rigging. Walter R. Mebane Jr., a professor of statistics and
political science at the University of Michigan at Ann Arbor, has been
analyzing precinct-level election results through the lens of *Benford's Law
*, a mathematical principle that describes how frequently particular digits
should appear in statistical tables of real-world data.

In a recent paper, "*Election Forensics: Vote Counts and Benford's Law*,"
Mr. Mebane suggested that in precincts with voting irregularities, the
digits in the election returns are not distributed in the way that Benford's
Law says they should be. He has detected such anomalies in returns from *Ohio's
2004 election*, Mexico's 2006 election, and Bangladesh's 2001 election.

If this theory pans out, then statisticians could routinely audit election
returns for violations of Benford's Law. But Mr. Mebane cautioned that the
idea needs further testing and development, and he stressed that certain
kinds of election tampering would not be revealed by his technique.

Susan D. Hyde, an assistant professor of political science at Yale
University, proposed that when international observers monitor an election,
they should sometimes use random assignment to choose which voting sites
they monitor, rather than concentrating on known trouble spots. (Ms. Hyde
designed such an experiment for the monitoring of Indonesia's 2004

True random assignment, Ms. Hyde said, can help reveal whether there is an
"observer effect." That is, does the presence of monitors deter malefactors
from ballot stuffing or voter intimidation? If the election returns in a
randomly selected monitored precinct are significantly different from the
returns in similar nearby nonmonitored precincts, that will be a sign of
possible wrongdoing.

Tracy A. Campbell, an associate professor of history at the University of
Kentucky, sketched the story of the decades-long effort to clean up a
corrupt electoral system in Louisville, Ky., in the late 19th century. He
stressed that Louisville's corruption involved graft and job patronage that
enveloped the entire city. "Election fraud involves more than just political
hacks or party officials," said Mr. Campbell, who is the author of Deliver
the Vote: A History of Election Fraud, An American Political Tradition,
1742-2004 (Carroll & Graf, 2005). "In many instances, voters were and are
willing accomplices."

A dissenting note was offered by J. Morgan Kousser, a professor of history
and social science at the California Institute of Technology. Mr. Kousser
agreed with his fellow panelists' skepticism about election systems -- but
he warned that the public should be equally skeptical of movements for
election "reform."

"There is a great American tradition of using allegations of fraud to
justify disfranchising people," said Mr. Kousser, who is the author of The
Shaping of Southern Politics: Suffrage Restriction and the Establishment of
the One-Party South, 1880-1910 (Yale University Press, 1974). He said he
feared that some potential voters would be driven away by proposed new
state-level requirements for photo identification.

More broadly, Mr. Kousser suggested that scholars and public advocates have
been too concerned with "technical" elements of election reform. State
governments should spend more money on training poll workers and less money
on buying new machines, he argued.

"We should pay more attention than we have been to nontechnical factors," he
said. "We need to keep our eye on protecting the fundamental right to vote."

Background articles from The Chronicle:
Georgia's Unusual 'Electoral College' (1/19/2007)

Should Election Science Become an Academic Discipline? One Professor Thinks
So (11/29/2006)

Hoping to Avoid Glitches With Electronic Voting Machines, One County Calls
in the Computer Scientists (11/9/2006)

Course Takes Students to El Salvador as Election Monitors (9/8/2006)

Experts Remain at Odds Over E-Voting (11/12/2004)

Motherhood, Apple Pie, and Election Fraud (12/10/2004)

Colleges Should Mobilize to Protect Voters' Rights (9/17/2004)

OVC-discuss mailing list
By sending email to the OVC-discuss list, you thereby agree to release the content of your posts to the Public Domain--with the exception of copyrighted material quoted according to fair use, including publicly archiving at http://gnosis.python-hosting.com/voting-project/
= The content of this message, with the exception of any external
= quotations under fair use, are released to the Public Domain
Received on Sun Sep 30 23:17:05 2007

This archive was generated by hypermail 2.1.8 : Sun Sep 30 2007 - 23:17:20 CDT